BENCHMARK EXPLORER
Compare image formats using matched measured quality rather than assuming that “quality 80” means the same thing in every encoder.
| Format | Estimated size | Saving vs JPEG |
|---|---|---|
| JPEG | 500.0 KB | 0% |
| WebP | 407.5 KB | 18.5% |
| AVIF | 305.0 KB | 39% |
This is a proportional visualization of the published savings, not a fresh encode of your file.
| Content type | WebP savings vs JPEG | AVIF savings vs JPEG |
|---|---|---|
| Photographs | 13–24% | 37–41% |
| Screenshots | 64% | 75% |
| Illustration | 46% | 66% |
These figures come from Rin Blog's published six-image matched-quality experiment, where outputs were searched to reach SSIM 0.98. They are reproduced here as an interactive reference, not presented as a new independent measurement.
JPEG, WebP and AVIF quality numbers are encoder-specific. Comparing q80 against q80 can therefore mix file size with different levels of visible degradation. A matched-quality test fixes a quality target first, then finds the smallest output that reaches it.
Content type matters too. Photographs, screenshots and flat illustrations stress codecs differently. A benchmark based on one photograph can therefore give misleading advice for a technical blog full of screenshots.
The source experiment used six images published on Rin Blog: four photographs, one screenshot and one flat illustration. Images were resized to the blog's publishing dimensions, encoded with Pillow, and compared using SSIM. Read the original methodology before treating the numbers as universal.